advertorch v0.1: An Adversarial Robustness Toolbox based on PyTorch

By Gavin Weiguang Ding, Luyu Wang, Xiaomeng Jin. 2019

200 citations recorded by the source corpus.

advertorch is a toolbox for adversarial robustness research. It contains various implementations for attacks, defenses and robust training methods. advertorch is built on PyTorch (Paszke et al., 2017), and leverages the advantages of the dynamic computational graph to provide concise and efficient reference implementations. The code is licensed under the LGPL license and is open sourced at this https URL .

Subfields: Artificial Intelligence, Molecular Biology, Safety, Robustness & Interpretability, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Matthias Hein
2. Quoc V. Le
3. Bernt Schiele
4. Alan L. Yuille
5. Yisen Wang
6. Xingjun Ma
7. Francesco Croce
8. Maksym Andriushchenko

In Molecular Biology

1. Francesco Croce
2. Matthias Hein
3. Maksym Andriushchenko
4. Kry Yik-Chau Lui
5. Ruitong Huang
6. Yisen Wang
7. Dongxian Wu
8. Shutao Xia

In Safety, Robustness & Interpretability

1. Matthias Hein
2. Yisen Wang
3. Francesco Croce
4. Maksym Andriushchenko
5. Gavin Weiguang Ding
6. Kry Yik-Chau Lui
7. Prateek Mittal
8. Shutao Xia

In Signal Processing

1. Matthias Hein
2. Kry Yik-Chau Lui
3. Ruitong Huang
4. Gavin Weiguang Ding
5. Francesco Croce
6. Maksym Andriushchenko
7. Yash Sharma
8. Mung Chiang

Citations in the corpus

• On the Sensitivity of Adversarial Robustness to Input Data Distributions (2018, methodology): “We use the PGD attack implementation from the AdverTorch toolbox (Ding et al., 2019).”
• Test-Time Training for Out-of-Distribution Generalization (2019, methodology): “Self-supervision has also been used on videos (Wang & Gupta, 2015; Wang et al., 2019).”
• Test-Time Training with Self-Supervision for Generalization under Distribution Shifts (2019, background): “The implementation of the adversarial attack comes from the codebase of Ding et al. (2019).”
• Batch Normalization is a Cause of Adversarial Vulnerability (2019, methodology): “We used the PGD implementation from Ding et al. (2019) with settings as below.”
• RobustBench: a standardized adversarial robustness benchmark (2020, background): “There are many libraries that focus primarily on implementations of popular adversarial attacks such as FoolBox [110], Cleverhans [104], AdverTorch [33], AdvBox [47], ART [98], SecML [92], DeepRobust [85].”
• Skip Connections Matter: On the Transferability of Adversarial Examples Generated with ResNets (2020, methodology): “For proper implementation, we use open-source codes and pretrained models for our experiments, e.g. , AdverTorch (Ding et al., 2019) for FGSM, PGD and MI, and source/target models from two GitHub repositories †‡§ for all models.”
• MMA Training: Direct Input Space Margin Maximization through Adversarial Training (2018, methodology): “For both (cid:96) ∞ and (cid:96) 2 PGD attacks, we use the implementation from the AdverTorch toolbox (Ding et al., 2019b).”